Counting the Dead

First, let me get this off my chest. What is the value of these articles, especially those purporting to show no more significant deaths in 2020 than 2019? Will that mean that COVID-19 is a hoax, perpetrated for some ulterior motive? Is there something to be gained by saying because the aggregate mortality numbers varied little from year to year, if in fact, they do differ so little, that we needn't lockdown and treat? Are the frail or susceptible who died of "COVID" instead of a heart attack treated fairly in a simple summary, or should we consider years of life lost and other economic measures? What I find particularly troubling about the obsession with these numbers is the hindsight gotcha. Everyone knows better in hindsight; the question of what, given the same circumstances and unknowns, we might have done differently is conjecture in the face of a novel disease.

Thank you for bearing with that moment. Now let's get to the issue underlying those mortality statistics, a topic covered previously here and here, death certification. Spoiler alert: statistics based upon death certification can undercount and overcount because the process is inherently subjective.

The individual certifying [1] a death

"performs the final act of care to a patient with a well-thought-out and complete death certificate that will allow the family to close the person's affairs."

As the guidance for physicians in NY State notes, the underlying cause of death "depends on the specific medical terms entered on the certificate and the sequence in which they are entered." Certification is a means of telling a story, describing a chain of events of HOW underlying causes resulted in an individual's death – a series of dots that can be connected in several ways.

The cause of death does not include final physiologic derangements, what attorneys might call the most proximate cause, e.g., the heart stopping (asystole) or too low a pH (acidosis) – these are terminal events. In the case of COVID-19 pneumonia resulting in an inability to oxygenate (hypoxia), hypoxia is a terminal event, not the cause of death.

The cause of death is "the final disease, injury or complication that directly led to the decedent's death." In the case I just described, the cause of death would be pneumonia. If I had a culture indicating a specific bacterial or viral source, I could be more specific. Viral pneumonia secondary to COVID-19.

The most reliable means of knowing the cause of death is an autopsy, a procedure performed on less than 10% of deaths in the US annually. A study from Norway demonstrated that autopsy altered the cause of death in 61% of cases; in half of those cases, the cause of death was an entirely different disease group, for example, changing from cardiovascular to malignancy. Signing a death certificate does not mean the statements are correct. Often, in cases outside the setting of acute care, we have to use the presence of other diseases or epidemiology to determine "the cause." Here is an example from a test question, "ripped from today's headlines."

"Scenario: A 68-year-old man was found dead at home. He had not seen a doctor in many years and has no known past medical history. The decedent's home is locked and secured, and there is no sign of illegal drug use or foul play. The decedent's family is at the scene, they do not want an autopsy performed and they have chosen a funeral home."

The correct response:

"Although the decedent has no past medical history and an autopsy was not performed, given the decedent's age the most likely cause of death is a cardiac event. Even though the certifier may feel uncomfortable as the cause of death is not known with absolute certainty, it is unacceptable to punish the family by calling the cause of death undetermined because they declined an autopsy." [emphasis added]

In the late winter and spring, would you fault a physician for writing COVID-19? Even in the early spring, the CDC allowed for writing presumed or probably COVID-19; positive cultures were not required. Like it or not, the truth is that our vital statistics on causes of death are fuzzy and have been made more ambiguous by the pandemic.

The ambiguity of judgment extends into the manner of death; natural, homicide, suicide, accidental, and undetermined. I mention this because of the opioid crisis - those deaths are handled a bit differently. Consider someone found to have died from a drug overdose; dead in their bedroom, a few pills scattered around along with a half-empty bottle of vodka. Is this a death due to drug abuse or acute and chronic substance abuse? It might be and would be coded as a death due to "mental health or behavioral issues." The correct coding is

"drug intoxication or drug toxicity which indicate a physical misuse of the substance. Using the correct terms will help to ensure that the death gets coded properly so that statistics about drug deaths are accurate."

Deaths related to drug intoxication are presumed to be due to "reckless behavior" and are classified as accidents. They are considered suicides when there is an "indication of intent of self-harm and death in order for the manner of death to be listed as suicide." The term "deaths of despair" often used in behavioral sciences to described drug overdoses refers to both those deaths deemed accidents or suicides.

Death certification involves judgment and is not as definitive as we would hope. Completion of death certificates is not a lesson learned in medical school; nor a specific training requirement of medical or surgical residency – it is a "skill-set" passed along from untrained senior residents to newbie interns. Hoping that the data we collect about deaths is as definitive as the deaths themselves is a fool's errand. You will find those numbers throwing more shadow, serving as a breeding ground for conspiracy, than light.

[1] Pronouncers are often depicted in dramatic TV moments, announcing the time of death; the paperwork of certification is left to another physician, the certifier, usually lower in the hierarchy, e.g., intern or resident.